Re: vacuum, performance, and MVCC

Ühel kenal päeval, E, 2006-06-26 kell 09:10, kirjutas Mark Woodward:
> > Ühel kenal päeval, R, 2006-06-23 kell 17:27, kirjutas Bruce Momjian:
> >> Jonah H. Harris wrote:
> >> > On 6/23/06, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> >> > > What I see in this discussion is a huge amount of "the grass must be
> >> > > greener on the other side" syndrome, and hardly any recognition that
> >> > > every technique has its downsides and complications.
> >> >
> >> > I'm being totally objective. I don't think we should abandon
> >> > PostgreSQL's overall design at all, because we do perform INSERTs and
> >> > DELETEs much better than most systems. However, I've looked at many
> >> > systems and how they implement UPDATE so that it is a scalable
> >> > operation. Sure, there are costs and benefits to each implementation,
> >> > but I think we have some pretty brilliant people in this community and
> >> > can come up with an elegant design for scalable UPDATEs.
> >>
> >> I think the UPDATE case is similar to the bitmap index scan or perhaps
> >> bitmap indexes on disk --- there are cases we know can not be handled
> >> well by our existing code, so we have added (or might add) these
> >> features to try to address those difficult cases.
> >
> > Not really. Bitmap index scan and bitmap index are both new additions
> > working well with existing framework.
> >
> > While the problem of slowdown on frequent updates is real, the suggested
> > fix is just plain wrong, as it is based on someones faulty assumption on
> > how index lookup works, and very much simplified view of how different
> > parts of the system work to implement MVCC.
>
> Yes, the suggestion was based on MVCC concepts, not a particular
> implementation.

On the contrary - afaik, it was loosely based on how Oracle does it with
its rollback segments, only assuming that rollback segments are kept in
heap and that indexes point only to the oldest row version :p

> > The original fix he "suggests" was to that imagined behaviour and thus
> > ignored all the real problems of such change.
>
> The original suggestion, was nothing more than a hypothetical for the
> purpose of discussion.
>
> The problem was the steady degradation of performance on frequent updates.
> That was the point of discussion. I brought up "one possible way" to
> start a "brain storm." The discussion then morphed into critisizing the
> example and not addressing the problem.

The problem is heatedly discussed every 3-4 months.

> Anyway, I think some decent discussion about the problem did happen, and
> that is good.

Agreed.

Maybe this _was_ the best way to bring up the discussion again.

--
----------------
Hannu Krosing
Database Architect
Skype Technologies OÜ
Akadeemia tee 21 F, Tallinn, 12618, Estonia

Skype me: callto:hkrosing
Get Skype for free: http://www.skype.com